Feature-preserving proxy mesh generation

Abstract

Systems and methods provide micro-credential accreditation. The systems and methods analyze, using one or more prediction models, received text submissions received from applicants via interaction with an applicant device. The prediction model(s) fit one or more micro-credentials to the received text submission, which may collectively or independently qualify the applicant for one or more accreditation credits. By processing the received text submission, the systems and methods allow for consistent and standard output of micro-credentials by the prediction model(s). Furthermore, the systems and methods provide for monitoring the prediction model output(s) to ensure ethical fairness across varying demographic groups of applicants.

What is claimed is: 
     
         1 . A method for micro-credential accreditation, comprising:
 receiving a text submission from an applicant device describing an event the applicant experienced;   processing the text submission with a predictive model to fit at least one micro-credential to the text submission, the micro-credential at least partially qualifying the applicant for credit from an accreditor;   outputting the at least one micro-credential.   
     
     
         2 . The method of  claim 1 , wherein outputting the at least one micro-credential includes providing school credit to the applicant. 
     
     
         3 . The method of  claim 1 , wherein outputting the at least one micro-credential includes displaying, on the applicant device, a chart indicating required micro-credentials to receive a desired accreditation credit. 
     
     
         4 . The method of  claim 1 , the receiving a text submission comprising providing one or more prompts to the applicant device, and receiving responses to said one or more prompts from the applicant device. 
     
     
         5 . The method of  claim 4 , the prompts including selectable prompts including a plurality of selectable options, and fillable prompts including an input box for receiving text input. 
     
     
         6 . The method of  claim 5 , further comprising processing the responses to generate a compiled submission, the compiled submission including a text string, portions of the text string corresponding to a selected one of the selectable options compiled into a first portion text string, and input text from the input box used as a second portion of the text string. 
     
     
         7 . The method of  claim 1 , further comprising generating a processed submission by:
 removing gibberish text within the text submission;   standardizing text within the text submission;   removing white space within the text submission;   removing stop-words within the text submission;   stemming individual words or phrases within the text submission; and,   performing lemmatization on the words or phrases within the text submission;   wherein said processing the text submission includes processing the processed submission.   
     
     
         8 . The method of  claim 1 , further comprising generating a processed submission by identifying a term frequency inverse document frequency (TD-IDF) matrix from the received text submission; wherein said processing the text submission includes processing the TD-IDF matrix. 
     
     
         9 . The method of  claim 1 , the processing the text submission comprising applying the text submission, or a processed version thereof, to a Pachinko Allocation Model to fit the at least one micro-credential. 
     
     
         10 . The method of  claim 1 , the processing the text submission comprising applying the text submission, or a processed version thereof, to a plurality of different prediction models each generating a list of fit micro-credentials, and implementing a consensus algorithm to develop a final list of one or more micro-credential from each least of fit micro-credentials. 
     
     
         11 . The method of  claim 10 , the consensus algorithm including one or both of Boosting and stacked generalization. 
     
     
         12 . The method of  claim 1 , further comprising monitoring previously output micro-credentials by demographic groups of applicants to identify whether the predictive model consistent between each demographic group; and updating the predictive model when the predictive model is not consistent within a fairness metric. 
     
     
         13 . The method of  claim 12 , wherein said monitoring comprises calculating a plurality of conditional statistical parity (CSP) values, each CSP value defining probability of a target micro-credential being issued to each demographic group; and comparing differences between CSP values to a CSP threshold. 
     
     
         14 . A system for micro-credential accreditation, comprising:
 a processor;   memory operatively coupled to the processor;   a micro-credential predictive module defined as computer-readable instructions within the memory and defining:
 a predictive model that, when the micro-credential predictive module is executed by the processor, fits a text submission to one or more target micro-credentials, and outputs the fit target micro-credentials as assigned micro-credentials. 
   
     
     
         15 . The system of  claim 14 , wherein the predictive model includes a plurality of target classifiers, each associated with one or more of the target micro-credentials. 
     
     
         16 . The system of  claim 14 , the target micro-credentials including a hierarchy of levels of micro-credentials. 
     
     
         17 . The system of  claim 14 , further comprising a model maintenance module that, when executed by the processor, identifies a fairness metric the one or more target micro-credentials. 
     
     
         18 . The system of  claim 17 , the fairness metric based on conditional statistical parity of the one or more target micro-credentials between different demographic groups. 
     
     
         19 . A method for maintaining a predictive model used for micro-credential accreditation, comprising:
 outputting a plurality of micro-credentials, said micro-credentials being fit, from a group of target micro-credentials, based on a prediction model analyzing a plurality text submissions received from one or more applicant devices;   analyzing a fairness metric of the target micro-credentials by determining a likelihood that each target micro-credential will be fit to each of a plurality of demographic groups;   updating the prediction model when the fairness metric indicates that the prediction model does not output the target micro-credentials across the demographic groups within a fairness threshold.   
     
     
         20 . The method of  claim 19 , wherein analyzing the fairness metric comprises:
 calculating a plurality of conditional statistical parity (CSP) values, each CSP value defining probability of a target micro-credential being issued to each demographic group; and   comparing differences between CSP values to a CSP threshold.